With the objective of addressing the scheme of the ght against poverty in Mexico and the de nition of strategies for an adequate public policy, an analysis methodology is proposed that identi ed the probability of risk in ve localities affected by poverty. For this, the degree of sensitivity of the variables that explain the poverty condition of the heads of household in 5 rural locations in Mexico was identi ed; Based on information collected through a census, a probabilistic regression model was formulated, with which three types of indicators were estimated: marginal effects, elasticities and odds ratio. Among the main results, it is worth highlighting that variables such as income, educational level, overcrowding and housing services have an inverse relationship in probability, while food, sex, and speaking indigenous language indicate a direct relationship in the prevailing factors. about the probability that the head of household is poor. Concluding that the explanatory factors of qualitative and quantitative nature of the head of household, in combination with those related to housing and the home, allow generating probabilitydistributions consistent with the theoretical argument.